REVIEW ARTICLE
Nanotribology: friction, wear and
lubrication at the atomic scale
Bharat Bhushan, Jacob N. Israelachvili & Uzi Landman
Friction, wear and lubrication between materials in contact are of fundamental importance in
many pure and applied sciences. Owing to the development of experimental and computer-
simulation techniques for studying these phenomena at the atomic scale, an understanding is
beginning to emerge of the molecular mechanisms of tribology in thin films and at surfaces.
UnpeRstaNpine the atomic processes occurring at the inter-
face of two materials when they are brought together separated
or moved with respect to one another is central to many techno-
logical problems, including adhesion, contact formation, fric-
tion, wear, lubrication, nanoindentation, fracture and
machining’ +, At most interfaces of technological relevance, con
tact occurs ai numerous asperities'. Consequently, the import-
ance of investigating single asperity contacts in studies of the
fundamental micromechanical and tribotogical properties of sur
faces and interfaces has been long recognized”. The recent
emergence and proliferation of proximal probes, in particular
tip-based microscopes (the scanning tunnelling microscope and
the atomic-force microscope) and the surface-force apparatus,
and of computational techniques for simulating. tip-surface
interactions and interfacial properties, has allowed systematic
investigations of interfacial problems with high resolution as well
as ways and means for modifying and manipulating nanoscale
structures. These advances provide the impetus for research
aimed at developing a fundamental understanding of the nature
land consequences of the interactions between materials on the
atomic scale, and they guide the rational design of materials
for technological applications. In short, they have led to the
appearance of the new field of nanotribology’®", which pertains
to experimental and theoretical investigations of interfacial pro-
cesses on scales ranging from the atomic and molecular to the
‘microscale.
‘The surface-force apparatus (SFA)'™™, the scanning tunnel-
ling microscope (STM)", the atomic-force and friction-foree
microscopes (AFM and FFM)", as well as the quartz micro-
balance technique”, are widely used in nanotribological studies.
The SFA was developed in the late 1960s and is commonly
employed to study both static and dynamic properties of molecu-
larly thin films sandwiched between two molecularly smooth
surfaces. The STM, developed in 1981, allows imaging of electr-
cally conducting surfaces with atomic resolution, and has been
used for imaging of clean surfaces as well a of lubricant
‘molecules. The introduction of the atomic-force microscope in
1985 provided @ method for measuring ultra-small forces
between a probe tip and an electrically conducting or insulating
surface, and has been used for topographical measurements of
surfaces on the nanoscale, as well as for adhesion and electro-
static fore measurements. Subsequent modifications of the AFM
led to the development of the friction-force microscope (FFM).
designed for atomic-scale and microscale studies of friction. This
instrument measures forces transverse to the surface. The AFM
is also being used for investigations of wear, indentation, detcc-
tion of transfer of material, boundary lubrication, and nanofab-
rication and machining, Meanwhile, significant progress in
understanding the fundamental nature of bonding and inter-
actions in materials, combined with advances in computer-based
‘modelling and simulation methods, have allowed theoretical
studies of complex interfacial phenomena with high resolution
in space and time'*"®, Such simulations provide insights into
atomic-scale energetics, structure, dynamics, thermodynamics,
transport and theological aspects of tribological processes.
NATURE - VOL 374 - 13 APRIL 1995
Furthermore, these theoretical approaches guide the interpreta-
tion of experimental data and the design of new experiments,
provide a framework for analysis and development of unifying
concepts, and enable the prediction of new phenomena based
‘on atomistic principles.
Here we review several of the principal theoretical and experi-
‘mental aspects of research in nanotribology. The nature of inter~
actions between (wo surfaces brought close together, and those
between two surfaces in contact as they are separated, have been
studied experimentally with the surface-force apparatus. This
hhas led to a basic understanding of the normal forces between
surfaces, and of the dramatic way in which these are modified
by the presence of an intervening medium such as a liquid or a
polymer ‘brush’ adsorbed or chemically grafted to the surfaces
The frictional properties of such systems have been studied by
‘moving the surfaces laterally, and such experiments have pro-
vided insights into the molecular-scale operation of lubricants
such 5 thin liquid or polymer films. Often, these lubricants
‘behave in ways that would not be predicted on the basis of their
bulk-scale behaviour alone.
Complementary to these studies are those in which the
AFM and FFM are used as a ‘model asperity’ in contact
with a solid surface. These experiments have demonstrated
that the relationship between friction and surface topography
is not always simple or obvious. AFM studies have also
revealed much about the nanoscale nature of intimate con-
tact—of indentation and wear. Computer simulations of tip
surface interactions at the atomic scale have shown the surface
need not behave as an elastic medium but can deform plast-
ically, for example leading to ‘necking’ between surface layers
and the probe tip.
Surface-force studies of thin films
‘The surface-force apparatus has long been used to measure both
static'™”” and dynamic!™**"* forces between smooth, macro-
scopic surfaces, As with AFMs and most other force-measuring
devices, forces are determined from the defection of a spring,
and displacements are measured using some optical, electrical,
capacitive or strain-gauge technique. The prineiple of operation
of SFAs (Fig. 1) is thus fairly simple; the challenge comes in
‘measuring very weak forces, sub-microscopie surface geometries
and surface separations at the dngstrém level.
Static (equilibrium) properties of thin liquid films. The inter-
molecular interaction potential or force function between (wo
surfaces interacting in or across a liquid medium ean be attrac-
tive, repulsive, ‘oscillatory’ or a more complex function of separ-
ation. Recent experiments and theoretical studies (mainly
computer simulations) have shown that the structure of the
liquid molecules and of the confining surfaces are very important
in determining what the force function will be® **. Furthermore,
two crystalline surfaces may induce a liquid to solidify (freeze)
when the gap thickness is close toa small multiple of the molecu-
lar diameter (Fig. 1, bottom left)” *°"*". The oscillatory force-
function resulting from such an interaction is illustrated in
Fig. 1 (top left)".
607REVIEW ARTICLE
FIG. 1 Mustration of the principle of the surtaceforce
‘apparatus (SFA) for measuring the normal forces, F, and
{ncuon or shear forces, f, between two smooth surfaces
‘of area A separated by a thin liquid fm of thickness O
(els 18-24), An optical interference technique allows the
Shapes of the surfaces and their separation, D, to Be
accurately monitored during foree and friction measure
‘ments. Typical contact area Ais 100 um”, and gap thick
esses can range from one molecular layer (D= a= 4 A,
where ois the molecular dismeter) to thousands of
ngstrdms. An oscillatory force function, FO}, ‘ypicaly
measured across simple liquids is shown at top lek, ane
The structure adopted by molecules in ultra-thin films is wer
erally very different from their bulk structure; other thin-film
properties may also be very different. The freezing point may be
higher or lower, depending on whether the ‘epitaxial interaction
of the confined molecules with the two surfaces is cooperative
or disruptive” ““* and molecular relaxation times can be many
orders of magnitude longer than in the bulk"? Tes also
interesting to note that even though one surface can have an
effect on the structure of the liquid molecules adjacent to
it", the effect of two confining surfaces is generally much
more dramatic, and can produce a greater variety of interfacial
(non-bulkclike) structures—amorphous, solid or liquid-crystal
line-—each of which gives rise (oa different type of force function
4s well as to different dynamic properties, as described below.
Dynamic and shear properties of thin films. The force between
two surfaces across a liquid or lubricant fluid tells us whether
two surfaces, when pressed together, will come into adhesive
contact or remain separated by-a thin, liqud-like layer"
This is crucially important for understanding the behaviour that
results when one of the surfaces is made to move laterally or to
shear past the other, as occurs during frictional and lubricated
sliding. For example, if the two surfaces are in true molecular
contact when sliding begins, their friction will be high and their
adhesion may be strong enough to tear them apart. If, however,
the surfaces are separated by one or more molecular layers of
the fluid, and if this cushioning film remains in the liguid-Hke
state during sliding, this may ensure that the friction force f will
‘be low and that sliding will proceed smoothly (Fig. 1, bottom
right). Experiments have shown that even one layer of water
molecules, approximately 0.25 nm thick, between two hydro-
philic surfaces can be sufficient to reduce the interfacial friction
force to a very low value”
On the other hand, ifthe surface-Auid interactions induce the
liquid molecules to solidify. the molecular configuration during
sliding will be much more complicated. In such cases the film
alternately melts and freezes during the motion’” “67 ®, as
shown schematically in the lower part of Fig. 1; the resulting
two common 'ypes of Trcton traces (ction force versus
time: f(2) are shown in the oscilloscopes at the bottom
fof the figure. During siaing, two smooth surfaces usualy
emain in one of the adhesive minima of the force func
tion, although the kinetic film thickness D, is usually
slighty diferent from the equilorium or state thickness,
>.
friction is of the ‘stick-slip’ type. Sticking occurs in the frozen
state, giving rise to the static friction force, f,. and slipping
‘curs in the shear-induced molten state giving rise tothe kinetic
friction force, f,. The real situation, even for the simplest sur-
faces and liquids, is actually much more complicated than shown
in Fig. 1: the direction of sliding is generally at some angle to
the surface lattices, the two surface lattices are not generally in
registry, the surface and liquid molecules have different diam-
eters, and they are generally not even spherical. This complexity
leads to complex force functions and friction-time relationships.
Even for quite complex force functions, however, recent experic
ments have shown that the forces of friction and adhesion are
closely related”
Shear-induced ordering transitions, Recent experiments and
computer simulations have shown that different types of ordered
structures may occur during frictional sliding’™*". The idea of
a ‘structure’ being induced by motion is an unfamiliar concept,
but ‘shear-induced ordering’ is at the heart of what goes on in
such thin films. Different types of ordered states can appear
depending on the temperature, load and sliding velocity, and
dynamic phase transitions between these states can also occur
during sliding, each giving rise to a complex stick-slip friction
pattern, Such effects are now believed to underlie many other
apparently unrelated dynamic processes involving. stick-slip
motion, such as earthquakes, the sound of a violin (and other
instruments), the squeaking of doors, and sensory perception
(of touch and even taste)"
Figure 2 shows some of the dynamic film structures that can
arise during the sliding of two surfaces that contain boundary
layers of surfactant molecules. In general, solid-ike films exhibit
stick-slip motion, amorphous-like films exhibit high friction due
{o the molecular interdigitations and entanglements occurring
between the two surfices, and liquid-tike films exhibit low, vis-
couslike friction with smooth sliding. These dynamic states are
not intrinsic properties of w shearing film (in the way that they
are for bulk condensed phases at any given temperature and
pressure): they depend not only on the temperature and pressure
NATURE » VOL 374 - 13 APRIL 1995REVIEW ARTICLE
‘Amorphous
regime
Fiction force ee
Solidike
regime
Liquide
regime
== Siding velocity, »
Temperature >
% u %
Salid “Amorphous
Liquid
$
ve Ss Sue
| Supariinetic siding]
i
FIG. 2 Schematic ustrations of aiferent types of dynamic thin-lm
structures that can arise during lubricated sliding. Top, dynamic phase
diagram representation of friction force (or rate of energy dissigation)
28 a function of temperature or siding velocty (adapted from refs 61
‘and 62). Under certain favourable coneitions involving surface grates
Chain molecules”, the entangled molecules at the interface may
undergo ‘shearinduced ordering. when they disentangle and become
‘aligned; at the same time the friction force drops signiianty (super.
kinetic friction),
‘of applied load, but also on the relative sliding velocity of the
two shearing surfaces (which is not necessarily the same at the
velocity of the drive: see Fig. 1). Higher sliding speeds make the
film more solid-ike, whereas slower speeds make it behave more
like a liquid. This principle, known as time-temperature super-
position, has recently been found to be very useful in understand>
ing the effects of time, temperature, load and other variables on
the complex molecular processes occurring at a shearing inter-
Face and the resulting friction forces"
Attaining desirable friction properties. The attainment of low
adhesion, low friction and low wear (surface damage) is usually
‘a desirable practical goal. But there are cases where the aim may
be to attain a constant friction force (with load or speed) oF
even a high friction, as occurs in the case of clutch fluids. By
judiciously choosing the right lubricant fluid and, in some cases,
by chemically grafting chain-like molecules such as surfactants”
co polymers" to surfaces (these are known as boundary lub-
ricants), itis becoming possible to control the frictional proper:
ties of moving components in machines and other devices. For
NATURE » VOL 374 » 13 APRIL 1995
‘example, the grafting of certain surfactant molecules at a certain
coverage to surfaces has been found” to produce films that can
undergo a dynamic first-order transition from a high friction
state to a low friction state (Fig. 2). The advantage of such 2
lubricant system is that no lubricant fluid is actually ever applied
‘or needed—the surfaces are self-lubricating in the sense that they
are processed right from the start to carry their own lubricant
layer with them; also, very small quantities of material are
needed so that material costs should not be prohibitive even for
fairly expensive lubricants. This is a good example of how recent
research in the field of molecular tribology is opening up pos-
sibilities for creating new types of lubricant systems.
AFM/FFM studies of tribological processes
Adhesion“ *, friction”, wear"! °°" and tubrica-
tion®®” * at ‘the interface between two solids with and with-
out liquid films have been studied using the AFM and FFM.
Surface roughness is routinely measured using the AFM™ *,
and both instruments have been used for measuring. clastic
plastic mechanical properties””""7*7""" *, At most solid-solid
interfaces of technological relevance, contact occurs at numerous
asperities' “'*""; a sharp AFM/FFM tip sliding on a surface
simulates just one such contact'*"*. For measurements of surface
roughness friction forces and nanoscale scratching and wear, a
microfabricated square-pyramidal SisN, oF silicon tip with a tip
radius ranging from 10 to SOnm is generally used"® at loads
ranging from 10 to 150nN, For measurements of microscale
scratching and wear, and for nanoindentation hardness measure-
ments and nanofabrication, a three-sided pyramidal single-
crystal naturai-diamond tip with a tip radius of about 100:nm
is generally used! at relatively high loads, ranging from 10 to
150 wN.
Surface roughness, adhesion and friction. Solid surfaces, ire
spective of their method of formation, generally contain surface
irregularities. When two nominally flat surfaces are placed in
contact, surface roughness causes contact to occur at discrete
contact points. Deformation occurs at these points, and may be
either elastic or plastic, depending on the nominal siress, surface
roughness and material properties. The sum of the areas of all
the contact points constitutes the real area of contact, and for
‘most materials at normal loads, this will be only a small fraction
of the area that would be in contact ifthe surfaces were perfectly
smooth, In general, this real area of contact must be minimized
to minimize adhesion, frietion and wear"
Characterizing surface roughness is therefore important for
predicting and understanding the tribological properties of solids
in contact. The AFM has been used to measure surface rough
ress on length scales from nanometres to mierometres"™* ", Sur-
face roughness most commonly refers to the variations in the
height of the surface relative to a reference plane”. Commonly
measured roughness parameters. such as rm.s. surface height
and peak-to-valley distance, are found to be scale-dependent for
any given surface. The topography of most engineering surfaces
is fractal, possessing a self-similar structure over a range of
scales, BY using fractal analysis one can characterize the rough
ness of such surfaces with two seale-independent fractal param
eters D and C. which provide information about roughness at
all length scales". ‘These wo parameters are instrument-
independent and are unique for each surface. D (generally rang-
ing from 1 (o 2) primarily relates to distribution of different
frequencies in the surface profile, and C to the amplitude of the
surface height variations at all frequencies. A fractal model of
elastic-plastic contacts" has been used to predict whether con-
tacts experience elastic or plastic deformation, and to predict
the statistical distribution of contact points
To study friction mechanisms on an atomic scale, a well char-
acterized, freshly cleaved surface of highly oriented pyrolytic
graphite (HOPG) has been studied by Mate e” al.” and Ruan
and Bhushan" ‘The atomic-scale friction force of HOPG
exhibited the same periodicity as the corresponding topography
609REVIEW ARTICLE
(Fig. 3a), but the peaks in friction and those in topography were
displaced relative to each other" (Fig. 3h). A Fourier expansion
of the interatomic potential was used to calculate the conserv-
ative interatomic forces between atoms ofthe FFM tip and thos
fof the graphite surface. Maxima in the interatomic forces in the
normal and lateral directions do not occur at the same location,
which explains the observed shift between the peaks in the lateral
force and those in the corresponding topography. Furthermore
the observed local variations in friction force can be explained
by variation in the intrinsic lateral force between the sample and
the FFM tip", and these variations may not necessarily occur
as a result of atomic-scale stck-slip processes!"
Frictional forces of HOPG have also been studied on micro:
metre scales", Local variations in the microscale friction
of cleaved graphite are observed, which arise from structural
changes that occur during the cleaving process". The cleaved
HOPG surface is largcly atomically smooth but exhibits line-
shaped regions in which the coefficient of friction is more than
order of magnitude larger. Transmission electron microscopy
indicates that the line-shaped regions consist of graphite planes
of different orientation, as well as of amorphous carbon. Differ-
ences in friction can also be seen for organic mono- and mult-
layer films”, which again seem to be the result of structural
variations in the films. These measurements suggest that the
FFM can be used for structural mapping of the surfaces. FFM
measurements can be used to map chemical variations as indica~
ted by the use of the FFM with a modified probe tip to map
the spatial arrangement of chemical functional groups in mixed
organic monolayer films”. Here, sample regions that had
stronger interactions with the functionalized probe tip exhibited
larger friction
Local variations in the microscale friction of scratched
LX tovearay 0) Fon
610
ificant, and seen to depend on the local
surface slope rather than the surface height distribution ®”"
Directionality in friction is sometimes observed on the macro:
scale; on the microscale this is the norm'®**”"-" This is because
‘most ‘engineering’ surfaces have asymmetric surface asperities,
so that the interaction of the FFM tip with the surface is depen?
dent on the direction of the tip motion, Moreover, during surface
finishing processes material can be transferred. preferentially
onto one side of the asperities, which also causes asymmetry
and directional dependence. Reduction in local variations and
in directionality of frictional properties therefore requires careful
‘optimization of surface roughness distributions and of surface-
finishing processes.
Table I shows the coefficients of friction measured for two
surfaces on micro- and macroscales. The coefficient of friction
FIG. 3 a, Grey-scale plots of surface topography tle)
{and friction profiles (right) of a1